Re: [PATCH] (IA64) Fix ugly __[PS]* macros in <asm-ia64/pgtable.h>

[Jamie Lokier]

David Mosberger wrote:
>   Jamie> Yes I have.  Quite thoroughly.
> 
> Then you should have noticed that drivers/char/mem.c is using PAGE_COPY.

That's an interesting bug in drivers/char/mem.c.  PAGE_COPY is wrong,
for archs with separate execute permission or write-only mappings.

The correct argument in drivers/char/mem.c should be vma->vm_page_prot.

>   Jamie> In theory the Alpha can do exec-only pages, but it's __[PS]*
>   Jamie> map always gives read permission when there's execute
>   Jamie> permission.  I'm not sure if there's a reason for that, or if
>   Jamie> it just historically copied the i386 behaviour (Alpha was the
>   Jamie> first port).
> 
> I know why: back in those days, GCC emitted code for nested C
> functions that assumed an executable stack.  Also, Linus wasn't
> terribly eager to turn off execute-permission on data/stacks.  Even on
> ia64 we started out that way, until I saw the error in my ways.

We're both wrong.  I misread the Alpha code: it does have exec-only
pages.  What it doesn't have is write-only.  And exec-only pages
aren't relevant to GCC's requirements, which used to be
read-implies-exec (exec-only breaks exec-implies-read).

-- Jamie
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/